QR code ordering with AI: how it works and what restaurants should check in 2026

What is QR code ordering?

QR code ordering lets restaurant guests scan a code on their table, open the menu in their phone’s web browser and send an order straight to the restaurant, without waiting for a server to take it. It goes further than a QR code menu, which only displays the menu, and asks less than app ordering, which needs a download and an account.

The QR, or Quick Response, code dates to 1994, when DENSO Wave released it as a code designed for high-speed reading (DENSO). In a restaurant, each table gets its own code, so an order arrives already linked to the right table.

Restaurants use table QR codes in three ways, from the least involved to the most:

• QR code menus. The code opens a digital copy of the menu. Guests still order from a server.
• QR code ordering. Guests browse, choose options and send the order from their phone; some systems also take payment.
• QR code ordering with an AI assistant. Guests can also ask questions in their own words, and the assistant helps them find dishes, check what a dish contains and follow their order.

How does an AI assistant change QR code ordering?

An AI assistant lets guests ask for what they want instead of scrolling through categories. It can suggest dishes that fit a diet, explain what a dish contains using the restaurant’s published information, help choose options and report how an order is doing. Staff still approve orders, so the assistant speeds up ordering without taking decisions away from the team.

A plain QR menu works like a printed one on a smaller screen: guests scroll, read and decide on their own. An assistant turns the menu into a conversation. A guest can ask “What’s vegetarian and not spicy?” or “Does the curry contain milk?” and get an answer drawn from the menu itself.

That only works if the assistant is grounded: it answers from the restaurant’s own menu, ingredient and allergen data, and says so when it does not know. A general-purpose chatbot that guesses is a liability in a dining room, especially for allergens, as the section on allergens below explains.

In Synthra’s Eatly, for example, the assistant answers from the restaurant’s menu and published ingredient information, guests can call staff from the same screen, and every order reaches the restaurant’s dashboard for approval.

How does QR code ordering with AI work, step by step?

A guest scans the table’s code, the restaurant’s menu opens, and the guest browses or asks the assistant questions. They choose options and send the order. The order appears on the restaurant’s screen, where staff approve it or reply if something can’t be done. The kitchen prepares it, and the guest can follow its progress.

1. Scan. The guest points the phone’s camera at the code on the table. The menu opens in the browser, tied to that table.
2. Browse or ask. The guest searches the menu or asks the assistant about dishes, diets, ingredients and options.
3. Choose and send. The guest picks the sizes and options the restaurant offers and sends the order.
4. Approve or reply. The order lands on the restaurant’s dashboard. Staff approve it, or tell the guest when a request can’t be made, such as a change the kitchen can’t do.
5. Prepare and follow. The kitchen prepares the order, and the guest can ask how it is doing or call a member of staff from the same screen.

The approval step is a design choice: some systems send orders straight to the kitchen. Keeping a person in the loop costs a few seconds, and it catches requests the kitchen can’t meet before the guest is left waiting.

What are the benefits for restaurants and guests?

Guests order when they are ready instead of when a server is free, and can ask questions without feeling rushed. Restaurants receive orders and customizations in writing, linked to the right table, and staff spend less time walking to tables just to take orders, which leaves more time for the service that needs a person.

• For guests: no wait to order or to add a dessert, answers to menu questions at their own pace, and options they can check before sending.
• For staff: fewer trips that only take an order, customizations that arrive in writing instead of being remembered, and time back for welcoming, advising and solving problems.
• For the restaurant: orders tied to tables, a record of what each guest asked for, and a menu that can be updated in one place.

Vendors often publish figures for higher spending or faster table turns. Treat them as claims to test in a pilot at your own restaurant, with your own menu and your own guests.

What are the drawbacks of QR code ordering?

Many guests still prefer a printed menu, not everyone has a smartphone or wants to use one at dinner, and a QR code can be tampered with. An AI assistant adds a risk of its own: a wrong answer, especially about allergens, is the restaurant’s responsibility. Each of these can be managed, but none should be ignored.

In a survey of 850 US adults by Toast, which sells restaurant ordering systems, 81% said they preferred a physical menu (Toast). QR ordering works best as an option guests can choose, with printed menus and ordering from staff still available.

Codes can be faked. The US Federal Trade Commission has warned that scammers cover real QR codes with their own, for example on parking meters, to send people to lookalike sites (FTC). Restaurants should check their table codes regularly and make sure the page that opens uses their own recognizable web address.

Phones are not equally easy for everyone to use. The ordering page should meet the Web Content Accessibility Guidelines (WCAG) 2.2, the W3C’s accessibility standard, so guests using screen readers, larger text or other assistive settings can order too.

Can an AI assistant answer allergen questions safely?

Only if it answers from allergen information the restaurant has published and checked, never from guesses, and sends anything uncertain to staff. Restaurants already have to provide allergen information in many places, and a business is responsible for what its chatbot tells customers, so the assistant must say it doesn’t know when the data isn’t there.

In the EU, Regulation (EU) No. 1169/2011 has required allergen information for non-prepacked food, including restaurant dishes, since 13 December 2014, and it covers 14 allergens. In Ireland, for example, restaurants must give it in writing, beside each menu item or in a clearly visible place (FSAI).

In the US, the 2022 FDA Food Code calls for written notification to customers of the major food allergens in unpackaged food, and sesame became the ninth major food allergen on 1 January 2023 (FDA). The Food Code applies where state or local regulators adopt it.

Responsibility for a chatbot’s answers stays with the business. In Moffatt v Air Canada (2024 BCCRT 149), British Columbia’s Civil Resolution Tribunal held that the airline remained responsible for all the information on its website, including what its chatbot told a customer, and ordered it to pay $812.02 (Deeth Williams Wall). A restaurant’s ordering assistant is no different.

• Keep one source of truth: ingredients and allergens entered and checked by the restaurant, which the assistant reads.
• Mark dishes whose allergen information is incomplete, so the assistant says so instead of guessing.
• Ask guests with severe allergies to confirm with staff, and keep calling staff one tap away.
• Review the assistant’s answers during a pilot before relying on it.

How is table ordering with AI different from AI drive-thru ordering?

A drive-thru assistant has to understand speech through a speaker in a noisy lane and get the order right from audio alone. At the table, the guest sees the order on screen before sending it, can correct it, and staff approve it. Errors are easier to catch, and nobody has to repeat themselves.

Voice AI at the drive-thru shows how hard ordering by speech alone can be. McDonald’s ended its automated order-taking test with IBM, which had run at about 100 restaurants since 2021; McDonald’s had aimed for order accuracy above 95% before any wider rollout (Nation’s Restaurant News).

Others are expanding. Wendy’s said in May 2025 that its FreshAI drive-thru ordering, then in more than 160 US restaurants, would reach more than 500 by the end of the year (Restaurant Dive).

What should a restaurant check before choosing a QR ordering system?

Check where the assistant’s answers come from, how staff approve and reply to orders, how customizations and allergens are handled, whether guests need an app, what happens when a guest prefers to order from a person, and how codes, data and payments are secured. Then run a pilot with your own menu before committing.

Run the pilot during real service, not a demo: a busy night, a large table, a menu change and an allergen question. Synthra’s Eatly is in partner preview and is being built to work with the POS and payment systems restaurants already use; restaurants can apply to become launch partners.

Where do tabletop assistants and robots fit in?

QR ordering puts the assistant on the guest’s phone. A tabletop assistant puts it on the table, where it is always there and can be spoken to directly. Service robots take the same intelligence onto the floor to carry orders. A restaurant can start with the phone and add the others later, on one platform.

Synthra builds all three steps: Eatly on the guest’s phone, T1, a tabletop assistant in development, and the U1 Series robots on the floor. See how the three fit together, and our guide to robots for restaurants and hotels.

Sources

1. DENSO celebrates 25 years of the QR code with prestigious award, DENSO, 2020-10-29
2. How Guests Really Feel About QR Code Menus (Survey Data), Toast
3. Scammers hide harmful links in QR codes to steal your information, US Federal Trade Commission, 2023-12-06
4. Web Content Accessibility Guidelines (WCAG) 2.2, W3C, 2024-12-12
5. Allergen Information for Non-prepacked Food, Food Safety Authority of Ireland
6. 2022 Food Code: sesame added as a major food allergen, US Food and Drug Administration
7. BC Tribunal Finds Air Canada Liable For Inaccurate Advice Given By Website Chatbot, Deeth Williams Wall LLP, 2024-03
8. McDonald’s is ending its AI drive-thru test with IBM, Nation’s Restaurant News, 2024-06-17
9. Wendy's to deploy drive-thru AI to over 500 restaurants this year, Restaurant Dive, 2025-05-02

Product and company names mentioned in this article are trademarks of their respective owners and are used only to identify those products. Synthra Robotics is not affiliated with, sponsored by or endorsed by them.